Hi folks.

I had an idea early on the week to record the flavor at the final stage of roasting so that I could stop the next roast at the most appropriate time. My first attempt didn't go very well but I hope to improve it with time.

I got me 3 trays. Two for cupping and one to place the beans straight out of the trier. I am planning to do two beans types each week (two cupping trays). The cupping trays are 3x4 and the bean tray is 4x6. The cupping trays are made of aluminum (safer for brewing?).



I can record the roast and place marks, so I thought everything was going to be OK. But trying to write things up in the computer (to make up the content of this thread) while roasting at the same time messed me up Anyway, I learned a few things in my first try.

I tried to duplicate my roasts with the heat gun + my little roaster. But I usually let the heatgun make BT higher.

two cups = 135+138 = 273 grams Ethiopia Amaro Gayo.
fan = 5
charge at 347F
coast with power off till 3 mins after charge ~240F
turn PID power on at 3 mins after charge and wait till change color (~300F)
HeatGun on 300F 7:50



At 1C I didn't know what to do. Once I powered down the gun, BT went down a little because my BT probe was heated by the HeatGun. The Q3 roaster is extremely consistent with a PID. But with a heatgun, the readings of BT read higher than normal. When I power off the heatgun, there is a period of adjustment and BT becomes normal again. I forgot to mark the end of BT on the computer (I only marked the beginning).

This roast had a long dry time. That's OK as long as the roast is not very long. I'll find out tomorrow how the flavor changed with time.

Another thing was that the roast did not have enough power to climb fast after 1C (loaded with 273 grams). That load would be OK if I had let the heat gun do the muscle but I pulled it off at 1C. With a lesser load and a higher ET it would have climbed much faster.

This is my end result for cupping tomorrow (starts top-left down, then next column)



The worst part was that I didn't write when the sample started accurately. Next time I'll be ready. I didn't have a proper stopwatch ready. Next time I'll be ready.

my BT was ~403F when I started taking samples (BT dipped a little after 1C)

(these readings are from the computer program)
seconds ET BT

767.203 542.3 403.7
772.203 541.3 405.0
777.203 539.6 406.3
782.203 538.7 407.6
787.187 539.0 408.9
792.187 540.2 410.4
797.187 541.1 411.7
802.187 541.0 412.9
807.187 540.0 413.6
812.187 538.4 413.7
817.203 537.0 413.4
822.187 536.5 412.7
827.187 535.6 411.7
832.187 533.7 410.4
837.203 531.3 409.1
842.203 529.2 408.1
847.203 527.9 407.0
852.203 527.4 405.9
857.187 527.6 405.0
862.187 528.2 404.2
867.187 528.6 403.4
872.187 529.3 402.7
877.187 530.5 401.9
882.203 532.4 401.2
887.203 535.0 400.4
892.203 537.7 399.9
897.203 539.3 399.3
902.187 540.5 398.6
907.187 541.8 398.0
912.187 543.7 397.3
917.187 545.3 396.7
922.187 546.2 396.3
927.203 545.4 395.9
932.203 544.4 395.5
937.203 543.8 395.2
942.187 544.1 395.0
947.187 544.1 395.1
952.187 542.9 395.2
957.187 541.0 395.4
962.187 539.5 395.5
967.187 539.0 395.9
972.203 539.3 396.2
977.203 539.7 396.8
982.203 538.8 397.3
987.187 537.2 397.7
992.187 536.3 398.2
997.187 536.3 398.8
1002.187 537.1 399.3
1007.187 538.2 399.8
1012.187 538.6 400.5
1017.203 538.3 401.1
1022.203 538.2 401.8
1027.203 539.0 402.4
1032.187 540.0 403.0
1037.187 541.2 403.6
1042.187 541.8 404.2
1047.187 541.5 404.7
1052.187 541.5 405.2
1057.187 542.3 405.8
1062.203 543.5 406.3
1067.203 544.5 406.9
1072.203 544.6 407.6
1077.187 543.8 408.3
1082.187 543.2 408.7
1087.187 543.3 409.2
1092.187 544.1 409.6
1097.187 544.3 409.9
1102.203 543.4 410.4
1107.203 541.8 411.0
1112.203 540.9 411.7
1117.203 540.8 412.4
1122.187 541.2 413.2
1127.187 541.5 413.9
1132.187 540.8 414.5
1137.203 539.5 415.0
1142.203 538.7 415.6
1147.187 538.8 416.3
1152.187 539.5 416.9
1157.187 540.0 417.0
1162.187 542.0 410.3
1167.203 544.2 401.4
1172.203 546.1 392.7
1177.203 547.4 387.3

I was hoping to match each sample taken to a BT. The goal was to add 20 seconds to obtain the next sample temperature.

Tomorrow I hope to learn how the flavor changed during the final phase of this roast. The value of this test is in the accuracy of the measurements (I don't expect much). The last samples tasted a little toasted. Maybe next time.

Cheers

I was expecting bad shots from the remaining beans from the experiment but they are good for espresso. The light roast forgave the very long finish phase (and long dry phase). The flavor was better than my past week darker roast of the same beans, which was more towards chocolaty. It does pull better at low temps ~195F (for espresso). Higher temps made it less flavorful. Hadn't it been for the sake of the experiment, I would have never let it sit for so long at that temp. Who would have known. Being only the first date after roast, I know how tricky these flavors can be.

After a break of espresso coffee, I'll do the cupping of the above Amaro Gayo beans. One of these days I'll have to learn how to "translate" the flavor of cupping to espresso without taking huge samples. The flavor of cupping to drip is straight forward. The sampling rate of the above beans for cupping was supposed to be 20 seconds apart.

Cheers

OK here are my results.



(I drilled a little hole in the top left corner of the cupping tray to mark the first sample)

I transferred equal amounts of beans (1.5 grams) into each cup. Then I calculated that for a proper brew I needed about 30 grams of hot water, which is about half the height of the cups. At the end I used more water than that but that did not present a problem.

My grinder performed really well for such small amounts. However, I think that next time I will use a finer setting of 7 instead of 8.5. A setting of 8.5 was maybe to coarse for the purpose of testing because made everything taste good, meanwhile a setting of 7 would have made the contrast come out better.



Anticipating that the aluminum tray would behave as a heat sink, I use a kettle water temp of 212F, which did the job fine.



The grounds floated at first and then they went down. At first, I found difficult to taste the differences, but after doing one round I could start seeing changes (even though the grind setting was coarse).

The adjectives that I gave to the 12 samples were:

1,2,3 some acidity
4 premature (bean flavor)
5 hollow
6 less premature
7 some light cereal mixed
8 neutral
9 sweetest?
10 less acidic
11 lowered toned
12 cereal/lower frequency flavor

I think using a coarse setting made it more difficult to distinguish the flavors. Next time I'll use a finer setting and a bigger sample of ground coffee.

I am not sure if I could use an electronic sweetness meter (glucose). It may be worth considering.

Conclusion: Due to the coarse grinder setting used, I was unable to tell what was best. However, I got a good idea of how the flavor changes during the last phase of the roast by comparing the results at the same time and looking for contrasts.

Hope next time I do a better job.

Cheers

Wow...

Just looking at these posts and not even attempting to read them makes my head hurt and raises a few questions, like:

Just how much time do you have on your hands?

And...

Have you been taking your medication?

That's harsh, but it would be good if you would summarize your results:
-- The graph and temperature chart are simply raw data, fit for an appendix, but unreadable by anyone except yourself. Decoding it requires detailed knowledge of your roaster, sensor placement and logging set up
-- The payoff of all your work is the twelve samples. And you just label them 1 - 12 for this post? You need to label each with the roast time and temperature it was pulled. Then people can get some sense of how taste correlates with roast

Also, the brownie pan is visually neat, and a good way to collect samples. But you must cup blind to reach reliable conclusions.

another_jim wrote:But you must cup blind to reach reliable conclusions.

Damn, now he needs to lose his eye sight too in addition to getting off his meds? Talk about suffering for your art.

Sorry, just could not pass it up.


But, yeah, this looks like neat idea. I like it. I would not call it obsessive. How else would you learn?

Thanks for the comments.

jeje. Believe it or not, I have my hands full this weekend with other stuff and I had to do this in a rush. Every now and then I take a break and come here. This is stuff to chill.

I'll wrap everything on Sunday evening, when I have more time. The experiment came out a little confusing because I was adjusting myself to the obstacles/mistakes/etc. Next weekend I'll redo it and I'll be more prepared. I may add another button next week to the program to record events (eventN+time(mm:ss)+BT) to match the samples taken. That would make things a lot easier. I could mark them on the graph too. I can later zoom in and take a pic of the final phase only.

Cheers

It's always nice to spend an hour on an expensive computer to do the work a sharpie will do in 2 seconds When you take the sample, just use a sharpie to write the time and temperature on the tray-cup. (This is also how you mark cup bottoms for blind tasting).

Hi Ed.

Just by learning criteria alone, the test should work.

Light roast are difficult to me. And I may have been confusing "premature" flavor with "light roast", not knowing what I was missing. I was lucky with the espresso result and learned I could hold longer (though it wasn't very light). This test could also help detecting the caramelization sweet spot. So it is a combination of learning/selecting.

I guess there are two goals:

1 I don't know first hand how the flavor progresses. With this I hope to learn by comparing the samples at the same time.

2 In theory, I could chose the flavor (previously choosing roast style-light/dark) by what I like most, by a criteria (sweetness,acidity), recommendation (maybe looking for a particular flavor someone talked about), etc.

Cheers